Advanced and complicated models like the ones often implemented in LUMET systems can be cumbersome to use in their raw form. Unlike simple regression models, the equations may look clumsy and the free parameters don't tell you very much unless you are very familiar with these kinds of nonlinear models. LUMET systems make it easy for you to utilise the models without your needing to understand the technology in detail. At the same time, this is no black box. If you want to see what the models are doing, a few clicks of the mouse will also show you what is going on inside.
Every LUMET system is customised for the users. The system will contain different software components which are suitable for the user, and will incorporate one or several empirical or semi- empirical models which will usually be used together. The picture below shows the prediction screen of one such LUMET system for a hot rolling mill. It contains three nonlinear models of mechanical properties based on the composition of the steel, and the thickness of the strips.
LUMET systems can implement, in principle, almost any kind of empirical models, but they are particularly designed for nonlinear models based on neural networks. The models might also be combinations of neural networks with other kinds of models. Plain linear regression models can also be implemented.
Over the years, we have understood better and better the facilities which make the system more valuable for the user, and therefore, the LUMET systems of today are a result of almost continuous improvement of its features over 10 years.
One general principle has been maintained from the beginning of its development. The LUMET systems are always extremely easy to use without a user's manual, and will take very little or no effort to get used to it. Today, people in industries are overwhelmed with new software of all kinds. LUMET systems are designed to make your life easier, and not add to the list of things you have to learn.